DeDoDe: Detect, Don't Describe -- Describe, Don't Detect for Local Feature Matching

TL;DR

DeDoDe introduces a novel approach for keypoint detection in 3D reconstruction by learning directly from 3D consistency, leading to improved performance over traditional descriptor-based methods.

Contribution

The paper proposes a new method that learns keypoints directly from 3D consistency using large-scale SfM data, avoiding proxy tasks and improving detection reliability.

Findings

Significant performance improvements on multiple geometry benchmarks.

Effective semi-supervised two-view detection for sparse points.

Decoupling keypoint detection from descriptors enhances downstream usability.

Abstract

Keypoint detection is a pivotal step in 3D reconstruction, whereby sets of (up to) K points are detected in each view of a scene. Crucially, the detected points need to be consistent between views, i.e., correspond to the same 3D point in the scene. One of the main challenges with keypoint detection is the formulation of the learning objective. Previous learning-based methods typically jointly learn descriptors with keypoints, and treat the keypoint detection as a binary classification task on mutual nearest neighbours. However, basing keypoint detection on descriptor nearest neighbours is a proxy task, which is not guaranteed to produce 3D-consistent keypoints. Furthermore, this ties the keypoints to a specific descriptor, complicating downstream usage. In this work, we instead learn keypoints directly from 3D consistency. To this end, we train the detector to detect tracks from large-scale SfM. As these points are often overly sparse, we derive a semi-supervised two-view detection objective to expand this set to a desired number of detections. To train a descriptor, we maximize the mutual nearest neighbour objective over the keypoints with a separate network. Results show that our approach, DeDoDe, achieves significant gains on multiple geometry benchmarks. Code is provided at https://github.com/Parskatt/DeDoDe